My Angkor Wat sunrise image
This is the first post on this site related to photography, and I thought I’d start with a relatively controversial one. I’ve had some really nice compliments on the Angkor Wat sunrise image that you can see pictured above (thank you for those!) I admitted in my most recent post (see here) that this image was in fact a composite image; so I thought I’d write a post detailing how I created this.
The morality of composition
The first thing I’ll say is that I’m not in the business of taking terrible photographs and then spending hours polishing turds, as it were. I say this up front, as in a moment you may think otherwise. However, most of the images on my site are barely edited at all – this one is the only composite image on the site at the time of writing. I went to the effort because frankly, I think the temple deserved a better bloody sunrise behind it than the one we woke at 4 AM to see.
You’ll note I refer to it as an image, as it is not a single photograph – it is a piece of imagery; created from different sources (all my own of course). I’ve done my best to make it as realistic as possible, in this case – but of course you can have a lot of fun creating weird and wonderful pieces of art with some of these techniques.
Now for the shock factor – the Angkor Wat sunrise image is composed of two separate photos; and the sky comes from a third. The two originals Angkor shots are pictured below. They were taken on a Nikon D750, which has an excellent full frame sensor; which is specifically good in low light. It may not seem it from first glance at these images, but bear with me.
These images were shot at ISO 1600, as the light truly was very low – owing to it being around 5:00AM and cloudy. The ISO was the film speed in analogue world, but is effectively the camera’s sensitivity to light – the higher the ISO, the more sensitive to light it is.
This ISO setting on many cameras would generate a lot of noise (pixels o colour which aren’t really there) – but that’s where the D750 outperforms a lot of cameras; Its photodetectors are fantastic; and as such can detect very low light energy levels.
Each image was taken with a shutter speed of 1/20 second. This is pretty slow & is borderline for needing a tripod to stop image blur – but the image stabilisation in the lens I’m using is quite good, so these were handheld shots – as I didn’t want to carry my tripod all day. I also had to hold the camera lot higher than a tripod would have allowed – to get over the crowds’ heads. The aperture was set to F7.1 to give a balance of the right amount of light, and sufficient depth of field to keep everything sharp, with focal length 40mm. I used a Nikkor 28-300mm lens – I love this lens – so versatile – but that’s another story.
Why these images?
Now as you can see; these photos are far from ideal – but I could see there was sufficient light captured to allow for semi-reasonable post-processing in Lightroom. Without at tripod, at this time of day, this was the best I would get without introducing noise, motion blur, or depth of field issues. After all, I didn’t want a bright image – it’s sunrise. I also figured I could create a silhouette shot if needs be.
I did take many other pictures that morning, and some had better lighting – but the composition of this combined shot made for a good overall image, so that’s primarily why I chose them. They also had fewer people in it than many of the others.
Note that there’s a lot of overlapping content between the two images. This is necessary to stitch together a good panoramic image of the Angkor wat sunrise later.
The editing begins
So the first step is to edit each of these, identically, in Lightroom. Now for those of you that aren’t into your photography; think of Lightroom as a digital Darkroom. It’s not magic and wizardry… It’s simply a digital means of editing the raw light data. In a dark room, you use chemicals to edit the outcome of the raw ‘data’ on the film – this isn’t much different – well it’s quite different – but you get the picture (pardon the pun).
The below two pictures are the edited versions; with some minor tweaks in Lightroom.
The adjustments made include;
- exposure adjustment;
- enhancement of shadows and black areas;
- slight relative reduction of highlights to bring the cloud brightness in line;
- Temperature adjustment on the White balance to enhance the greens (make the picture warmer)
- Enhanced vibrance and saturation to add clarity to some of the stone work and greenery
- Introduced tone curve to add some contrast & remove black & white clipping
- No noise reduction; as I didn’t want to lose detail
- Lens profile correction – so that curvature of the landscape & any vignetting could be removed prior to any panoramic stitching.
- Dehaze – great tool – took some of the greys out and added further clarity to the now brighter image.
I will in future delve into detail on some of these functions of Lightroom, but they’re listed above in the order you see them in Lightroom for those that use it.
Creating the Panoramic
Clearly the above two images are still two separate images. I stitched these together in Photoshop (PS); which can be triggered in Lightroom by selecting the photos you want to merge, right clicking; and going ‘Edit In > Merge to Panorama in Photoshop’.
Now it is possible to merge to panorama within Lightroom and it normally does a great job. The reason I didn’t in this case was because unfortunately Lightroom isn’t quite as clever as PS. If you see the screenshot below, the Lightroom panoramic looks great – but you’ll notice it has a curve around the edges – because naturally as you rotate the camera to take different shots of a panoramic; you’re shooting a cylindrical image; which Lightroom then flattens for you. Unfortunately when I crop the Lightroom panoramic in order to remove the white oval shape; I lose the top of the temple in the pond reflection – which simply won’t do!
Why Photoshop for this stitch?
That’s where PS comes in; PS has a function called ‘Content Aware, fill transparent areas’ (see check box in below screen shot). This is awesome. It will create the panoramic the same as Lightroom (albeit it can cope with more complex panoramics); but it will then try and fill the white areas around the curve seen above, by looking at the nearby content & creating image data to fill the areas.
The result is seen below. Now, as you can see, it’s not perfect. It has created a bit of tree that I’ve circled in red; that clearly doesn’t exist, as it’s part of the reflection in the pond – but not part of the tree that’s being reflected. However, that’s easily removed using the Clone Stamp tool. The same goes for the various people in the photo. Careful use of the clone stamp tool can replace people with similar patches of nearby grass, or pillars in the structure etc… Very handy!
But where’s the friggin sunrise?
The sky in the above images was rubbish – let’s face it. Nothing we can do about that. We were only in Cambodia for a few days; and this was out last – so no more sunrise opportunities for proper shots. However I do know where the sun rises at Angkor – just behind the centre-right of the main structure if sat where I was. And thankfully Australia offered me some beautiful sunrises which would serve as beautiful Cambodian sunrises – I think anyway.
So the below was the sunrise image I chose and used for this image. Taken on Nobby’s beach in Australia.
Merging into the Angkor Wat shot
In order to merge this into the Angkor Wat shot, I opened this as a new layer in my Photoshop panoramic document; and put it to the bottom of the Layer pile so that it was the background layer. As it’s a single image, it’s not as wide as the one I’m trying to create, but using free transform, you can easily adjust the width of the image to fit. It stretches the pixels which isn’t ideal, but as long as you’re not stretching the image too much, you can get away with it.
Remove the crappy sky
It’s then time to set to work removing the sky from the original image, so that this background layer could be revealed. This is quite time-consuming so you have to be fairly committed.
The first step is to select your foreground layer (the Angkor picture in this case), and then I used the ‘Select & Mask’ function (in the select menu). This allows you to select areas of similar colours by clicking around. You can adjust the sensitivity of the tool to suit. You can see the settings I used. As the clouds were of a more consistent colour; I used those to carry out the selection, and then simply inverted the selection within the select & mask function on the right side of the screen. This process does the majority of the work, but naturally it leaves some sky behind. The below image shows the screen before selecting ‘Invert’ – the sky has all been selected & the rest of the image is masked. Once inverted; this effectively puts a mask over the sky – so that you can see straight through it to any layers behind.
The image below shows you how it looks with the old sky removed; and the new sky in the lower layer visible. It’s zoomed in to show you there are areas between the trees that still have the old sky present.
Getting into the detail
To fine tune the sky removal, I used the ‘erase background’ tool. This is a fairly intelligent tool, which when adjusted appropriately, will erase all similar colours to the one you click on, within the size of the brush you’re using. So if you have a 500 pixel circle covering the whole tree below, and the central cursor (in the middle of the circle) on an area of sky – it will erase all similar colours and in theory leave the tree behind. It takes some messing around with to get the settings right for your picture.
However eventually you end up with reasonable lines, that aren’t perfect close up -but are certainly good enough for some sized prints or web imagery. See below.
I then had to repeat all of the masking tasks on the reflected sky in the pond (grr!!), and ended up with an image that looked like the below. Now to me, it was apparent that this was two images pushed together; as the colouring was different between the two. So this is where adjustment layers come in.
Making the Angkor Wat sunrise real
In order to turn the image above from one that was clearly comprised of two photoshoots (with different colouring); I added in a number of adjustment layers that affected the Angkor Wat foreground layer & both layers together to try and bring them together. I adjusted the colour balance of the foreground layer to enhance the yellows and reds slightly more. Then added layers to adjust joint exposure; colour balance of whole image, and the brightness / contrast of the whole image. This gave the end result below. The differences are subtle. Please note that these are heavily compressed web-friendly images – so they won’t look perfect zoomed in or enlarged… The full resolution file is 8536 x 3912 pixels (33.4MP).
So that’s it – 300 man hours of effort for one picture. Nah not really – 299 at most – certainly for a practiced hand (not mine) this wouldn’t be a whole lot of work – maybe a couple of hours; even though it seems like more. But I think it’s delivered a fairly decent image of Angkor Wat sunrise; which sadly we didn’t get to see in as much glory as pictured here.
However, there’s a different part of my psyche that’s even happier with having created this picture without it actually having existed.